Each passing day of the Russian invasion in Ukraine is taking a toll on not only on the lives but the future livelihood of millions. At least 1/3rd of its population (over 12 million ) have migrated since February, 2022, most of its essential services are halted while schools and hospitals are being bombed. By a World Bank estimate Ukraine’s economy to shrink by 45%

While efforts are being made to address the first order issue of stopping the invasion, it is also important to start paying attention to what Ukraine might look like after the war and take a stock of the extent of loss in infrastructure, natural-resources and identify worst hit regions. Doing so, will not only help in targeting focused reparative efforts but also save time in policy response till surveys and ground assessments are undertaken and collected.

In the absence of real time data, satellite data can be a handy proxy Imagery from daytime satellites have already proven to be useful in preventive mobilization before the invasion started encroachment by Russian army, learning of Russia strategically targeted Ukrainian grains. Data from nightlights can further help in assessing the extent of economic disruption from loss in electrification, demolition of private property and public infrastructure.

Assessing impact of war in Ukarine from nightlights

The following maps show the drastic drop in nightlights across Ukraine from January to May 2022.

It can be seen that light intensity in highly-lit (bright yellow) regions, corresponding to cities including Kyiv, Lyiv, Kharkhiv have drastically reduced since January and continue on a downward trend. Milder yellow/orange regions potnetially capturing street lights from highways have almost disappeared, matching with ongoing narrative from the ground. Lights from dark blue regions capturing small settlement have almost disappeared.

Checking confounders I (Blackouts): Electricity Demand

Its highly likely that the such drastic reduction in light could be driven by a.) cloud-cover biophysical, seasonal noise and b.) strategic blackouts to stave off air-strikes. Assuming that the blackouts were not consistent across all the months, checking with electricity demand could potentially help in account for this. Following is chart of monthly electricity demand estimated by the IEA

source

It can be seen that the drop in lights match with the drop in monthly electricity, with a stark drop in May 2022 onwards and marginally decreasing levels after April. This seems to confirm the perception, direction, and magnitude of the fall. It can also be seen that the drop in lights is pronounced in eastern part of the country coinciding with Possibly implying that the drop in lights was not just due to cloud cover or blackouts but largely from loss in infrastructure and breakdown of economic activity.

Checking confounders I (Seasonal Noise): Compare with 2021 Data

Monthly nightlights map

Drop in nightlights across months

Caveat: Limitation on inferences to be drawn from nightlights

NOTE - Lights are not equivalent to income. While several studies have used earlier variant of Nightlights (DMSP) as a proxy for assessing economic growth, there are only a few studies finding so using latest nightlights product (VIIRS) - Nightlights map can be misleading with different color labels, the labels used here are based use the same value thresholds across all maps (Nightlight measures radiance of light in Watt/cm2/sr) - Nightlights data for month of June seems to been biased by external noise and is not included here